【Python】Mac Python安装liblinear机器学习库

亦凉 2022-05-28 09:15 288阅读 0赞

## 前言 ##

今天刚接触这个软件包，估计未来几天都会和它打交道了，会把相关的代码整理出来，持续更新。

## 安装 ##

liblinear是一个简单的求解大规模规则化线性分类和回归的软件包。它最讨人喜欢的特点就是速度快，和libsvm一样，都是由台大Chih-Jen Lin开发的，提供了多种编程语言API，下面简介安装流程。  
**下载**  
官网：[http://www.csie.ntu.edu.tw/~cjlin/liblinear/][http_www.csie.ntu.edu.tw_cjlin_liblinear]  
**解压**  
解压压缩包，移动到某个文件夹下（我习惯移到pip安装文件夹下，/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/）  
**Make**  
进入liblinear主目录，在terminal中执行make

$ make
    c++ -Wall -Wconversion -O3 -fPIC -c -o tron.o tron.cpp
    tron.cpp:119:35: warning: implicit conversion changes signedness: 'int' to 'unsigned long' [-Wsign-conversion] memcpy(w_new, w, sizeof(double)*n); ~^ tron.cpp:160:36: warning: implicit conversion changes signedness: 'int' to 'unsigned long' [-Wsign-conversion] memcpy(w, w_new, sizeof(double)*n); ~^ 2 warnings generated. ******中间的省略 6 warnings generated.

只有warnning，没error就OK。之后进入./python的文件夹下，在terminal中执行make

$ cd ./python
    $ make
    make -C .. lib
    if [ "Darwin" = "Darwin" ]; then \
            SHARED_LIB_FLAG="-dynamiclib -Wl,-install_name,liblinear.so.3"; \
        else \
            SHARED_LIB_FLAG="-shared -Wl,-soname,liblinear.so.3"; \
        fi; \
        c++ ${SHARED_LIB_FLAG} linear.o tron.o blas/blas.a -o liblinear.so.3

**配环境**  
拷贝./python的文件夹目录，将目录添加至python解释器的path中，执行：

$ export PYTHONPATH=/Users/username/Downloads/liblinear-2.01/python:$PAYTHONPATH

5.进入python， 运行import liblinear:

$ python
    Python 3.6.4 (v3.6.4:d48ecebad5, Dec 18 2017, 21:07:28)
    [GCC 4.2.1 (Apple Inc. build 5666) (dot 3)] on darwin
    Type "help", "copyright", "credits" or "license" for more information.
    >>> import liblinear
    >>> exit()

不报错就装好咯～

另外，找到了一个叫做pyliblinear的库，说是python的API，链接：[http://opensource.perlig.de/pyliblinear/index.html][http_opensource.perlig.de_pyliblinear_index.html]

### 几个相关的链接 ###

[https://blog.csdn.net/u010159842/article/details/49305031][https_blog.csdn.net_u010159842_article_details_49305031]  
[https://blog.csdn.net/sadfasdgaaaasdfa/article/details/48825875][https_blog.csdn.net_sadfasdgaaaasdfa_article_details_48825875]  
[https://blog.csdn.net/zouxy09/article/details/10947323/][https_blog.csdn.net_zouxy09_article_details_10947323]

[http_www.csie.ntu.edu.tw_cjlin_liblinear]: http://www.csie.ntu.edu.tw/~cjlin/liblinear/
[http_opensource.perlig.de_pyliblinear_index.html]: http://opensource.perlig.de/pyliblinear/index.html
[https_blog.csdn.net_u010159842_article_details_49305031]: https://blog.csdn.net/u010159842/article/details/49305031
[https_blog.csdn.net_sadfasdgaaaasdfa_article_details_48825875]: https://blog.csdn.net/sadfasdgaaaasdfa/article/details/48825875
[https_blog.csdn.net_zouxy09_article_details_10947323]: https://blog.csdn.net/zouxy09/article/details/10947323/